On the gap between the quadratic integer programming problem and its semidefinite relaxation

摘要

Consider the semidefinite relaxation (SDR) of the quadratic integer program (QIP): where Q is a given symmetric matrix and D is diagonal. We consider the SDR gap We establish the uniqueness of the SDR solution and prove that if and only if γ r :=n −1max{x T VV T x:x ∈ {-1, 1} n }=1 where V is an orthogonal matrix whose columns span the (r–dimensional) null space of D−Q and where D is the unique SDR solution. We also give a test for establishing whether that involves 2 r −1 function evaluations. In the case that γ r <1 we derive an upper bound on γ which is tighter than Thus we show that `breaching' the SDR gap for the QIP problem is as difficult as the solution of a QIP with the rank of the cost function matrix equal to the dimension of the null space of D−Q. This reduced rank QIP problem has been recently shown to be solvable in polynomial time for fixed r.